Types of studies

We considered randomised controlled trials that compared conservative (non‐endoscopic) treatments for impacted oesophageal food boluses against other conservative (non‐endoscopic) treatments, placebo, or watchful waiting.

Types of participants

Types of interventions

We considered the following:

• all non‐endoscopic treatments, as outlined above. These are broadly divided into enteral treatments, parenteral treatments, and watchful waiting;

• combinations of treatments, providing the allocated management did not include endoscopy;

• all doses and dosing regimens;

• all relevant parenteral routes: subcutaneous, intramuscular, and intravenous;

• all comparisons of the above treatments, including comparisons with placebo.

Types of outcome measures

Electronic searches

We searched the following electronic databases to identify relevant studies. We last conducted the searches on 18 August 2019.

• Cochrane Central Register of Controlled Trials (CENTRAL) (searched via Ovid; up to Issue 7, 2019) (Appendix 2).

• MEDLINE (searched via Ovid; 1946 to 18 August 2019) (Appendix 3).

• Embase (searched via Ovid; 1974 to 18 August 2019) (Appendix 4).

• CINAHL (Cumulative Index to Nursing and Allied Health Literature) (searched via EBSCO; 1982 to 18 August 2019).

Searching other resources

We searched the reference lists of the articles retrieved, including those of review articles. We attempted to contact authors of relevant studies to ask them to identify any other published and unpublished studies. We searched for errata or retractions of eligible studies at www.ncbi.nlm.nih.gov/pubmed on 21 August 2019.

Selection of studies

Two review authors (JH and NS) independently screened the titles and abstracts of studies identified by the above search strategy. We identified and excluded duplicates, and then coded unique articles as either 'retrieve' (eligible or potentially eligible/unclear) or 'do not retrieve' if clearly irrelevant. For articles coded as 'retrieve', we obtained the full‐text publication. Two review authors (JH and NS) then independently assessed the text to identify studies for inclusion. We excluded ineligible studies, recording the reasons for exclusion in the Characteristics of excluded studies tables. Any disagreements were resolved through discussion or in consultation with a third review author (JS). Had we identified multiple reports of the same study, we would have collated these, so that each study, rather than each report, formed the unit of interest in the review. The selection process is recorded in a PRISMA flow diagram (Figure 1).

Figure 1

---

---

PRISMA flow diagram of study selection process.

Data extraction and management

We used a standard data collection form for study characteristics and outcome data that was pre‐piloted and adapted as required. Two review authors (JH and NS) independently extracted study characteristics. We attempted to extract the following study characteristics.

• Methods: study design, total duration study and run‐in, number of study centres and location, study setting, withdrawals, date of study.

• Participants: N, mean age, age range, gender, severity of condition, diagnostic criteria, baseline lung function, smoking history, inclusion criteria, exclusion criteria.

• Interventions: intervention, comparison, concomitant medications, excluded medications.

• Outcomes: primary and secondary outcomes specified and collected, and time points reported.

• Notes: funding for study, notable conflicts of interest of study authors.

Two review authors (JH and NS) independently extracted outcome data from the included studies (for details see Characteristics of included studies). No outcome data were reported in an unusable way. Any disagreements would have been resolved by consensus or by involving a third review author (JS); this was not required as there were no disagreements. One review author (JH) entered the data from the data collection form into the Review Manager 5 file (Review Manager 2014). A second review author (NS) verified this for accuracy.

Assessment of risk of bias in included studies

Two review authors (JH and NS) independently assessed risk of bias using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). There were no disagreements between review authors requiring arbitration by a third review author. We assessed bias for the following domains.

• Random sequence generation.

• Allocation concealment.

• Blinding of participants and personnel.

• Blinding of outcome assessment.

• Incomplete outcome data.

• Selective outcome reporting.

• Other bias.

We graded each potential source of bias as either high, low, or unclear and have provided quotes from the study report, together with a justification for our judgement, in the 'Risk of bias' table. Had information on risk of bias related to unpublished data or correspondence with a trial author, we would have noted this in the 'Risk of bias' table. When considering treatment effects, we assessed the risk of bias for the studies that contributed to that outcome.

Measures of treatment effect

We analysed dichotomous data as risk ratios. We would have reported continuous data as mean difference or standardised mean difference. We would have summarised time‐to‐event outcomes using hazard ratios. We would have undertaken meta‐analysis if this were meaningful, that is if the treatments, participants, and the underlying clinical question were similar enough for pooling to make sense.

We did not encounter any reporting of medians and interquartile ranges, which may have suggested skewed data, but would have reported this and considered its implications.

Had we encountered multiple study arms in a single study, we would have included only the relevant arms. If the same meta‐analysis needed two comparisons, we would have halved the control group to avoid double counting.

Unit of analysis issues

We identified no non‐standard parallel group random allocation study designs. Had we identified cross‐over studies, we would only have considered the results from the first phase of the study, prior to cross‐over.

Dealing with missing data

We attempted to contact study investigators to clarify key study characteristics and to obtain missing outcome data. Where appropriate, we would have followed intention‐to‐treat principles.

Assessment of heterogeneity

We would have used the I² statistic to measure heterogeneity amongst the studies in each analysis (Higgins 2003). In the case of substantial heterogeneity, we would have performed subgroup analysis.

Assessment of reporting biases

Had we identified more than 10 studies for any individual parameter, we would have produced a funnel plot to explore potential publication bias. Sensitivity analysis for missing data was not required.

Data synthesis

For quantitative analysis, we would have performed a meta‐analysis using Review Manager 5, Review Manager 2014, and a random‐effects model (Deeks 2011).

Subgroup analysis and investigation of heterogeneity

Had it been applicable, we would have performed subgroup analysis of the type of conservative treatment, grouped by substance type (irrespective of dose, regimen, or route) or watchful waiting, with the aim of determining if any single enteral or parenteral treatment outperformed another, including watchful waiting. We would have used only the primary outcome measures.

Sensitivity analysis

Had it been required, we would have performed a sensitivity analysis including only studies at low risk of bias for all domains outlined in the Assessment of risk of bias in included studies section.

Summary of findings and assessment of the certainty of the evidence